Packages

• Pakete (installiert und) ladt
  • tidyverse
  • languageR
  • ggthemes
  • patchwork

# in the CONSOLE: install packages if needed
install.packages("tidyverse")
install.packages("languageR")
install.packages("ggthemes") # for customising our plots
install.packages("patchwork") # plot layouts

# Pakete laden
library(tidyverse)
library(languageR)
library(ggthemes)
library(patchwork)

1 Data frames

• data frames are a collection of variables, where

  • each variable is one column
  • each row is a single observation/data point
  • each cell in a row is linked

• data frames are just like spreadsheets, but are rectangular

• different words for data frames:

  • data frame
  • dataset
  • tibble (in the tidyverse)

Talking about datasets

• when we talk about our data, we use certan words to refer to different parts:

• a variable: a quantity, quality, or property you can measure

• a value: the state of a variable when you measure it

• an observation: set of measurements made under similar conditions

  • will contain several values each associated with a variable
  • an observation for a single variable is sometimes called a data point

• tabular data is a set of values, each associated with a variable and an observation

  • tabular data is tidy if each value is placed in its own cell, each variable in its own column, and each observation in its own row

Categorical and continuous variables

• how we visual the distribution of a variable depends on what type of data it represents: categorical or numerical
• a variable is categorical if it can take a small set of values that can be grouped together
• e.g., old/young, short/tall, grammatical/ungrammatical, L1/L2-speaker
• a variable is numerical (i.e., quantitative) if it can take on a wide range of numerical values
  • and it would make sense to add, subtract, compute the mean, etc.
  • can be continuous (decimal points make sense, e.g., 1.5cm)
  • or discrete (decimal points do not make sense, e.g., 1.5 children doesn’t make sense)
  • age, height, reaction times, format frequencies
• we produce different plots depending on what type of variables we want to visualise

2 Lexical Decision Task

• our first dataset contains data from a lexical decision task (LDT)

• in the LDT, participants press a button to indicate whether a word is a real word or a pseudoword

LDT variables

• common variables collected in a lexical decision task experiment are:
  • reaction time
  • accuracy (correct/incorrect)
  • word category (e.g., real/pseudo, noun/verb)
  • word frequency
• additional variables that might be collected could be:
  • participant demographics (e.g., age, L1/L2, gender)

3 lexdec dataset

• languageR is a companion package for the textbook Baayen (2008)
  • contains linguistic datasets, e.g., lexdec
• the lexdec dataset contains data for a lexical decision task in English
  • we will be working with variables such as reaction times and accuracy

lexdec variables

• a list of some of the variables is included in Table 1

LDT research questions

• before we conduct an experiment, we have research questions that we want to answer with the data
  • today we’ll address the following question:
    • do the reaction times differ between native and non-native speakers?

Load the data

• our data is available in the lanaugeR package we’ve already loaded
  • to print the data, just type the name of the dataset and run it
• below we only see a few variables, but you should see more in your console

lexdec

  Subject       RT Trial Sex NativeLanguage Correct PrevType PrevCorrect
1      A1 6.340359    23   F        English correct     word     correct
2      A1 6.308098    27   F        English correct  nonword     correct
3      A1 6.349139    29   F        English correct  nonword     correct
4      A1 6.186209    30   F        English correct     word     correct
5      A1 6.025866    32   F        English correct  nonword     correct
6      A1 6.180017    33   F        English correct     word     correct

• how many variables do we have? observations?

Save data as an object

• to save the data in our Environment, we have to assign it a name
  • let’s call it df_lexdec, which means “data frame lexical decision”

df_lexdec <- lexdec

• now we see it in our Enrivonment
  • double-click on it to view it in the Editor pane

Relevant variables

• Among the variables we have are:
  1. Subject: participant ID
  2. RT: logged reaction times
  3. NativeLanguage: the native language of the participant
  4. Word: what word was presented
  5. Class: if the word was animal or plant

Ultimate goal

• our ultimate goal today is to produce the following visualisation of the data
  • the plot shows the distribution (count) of reaction times and native language of the participants

4 Creating plots with ggplot2

• the tidyverse is a collection of packages that facilitate data tidying and visualisation
  • when we load the tidyverse, this collection of packages is automatically loaded
• the ggplot2 package is a tidyverse package that builds plots in layers

Layer 1: empty canvas

• the first layer with the function ggplot() is like an empty canvas

ggplot(data = df_lexdec)

Layer 2: mapping aesthetics

• next we tell ggplot() how to visually represent our variables
  • we use the mapping argument of ggplot() to define how variables are mapped to visual properties (aesthetics)
• for our first plot, we want to map reaction times (RT) on the x-axis (the bottom of the plot)
  • we wrap the logged RT in the exp() function to get RTs in milliseconds (for reasons we won’t discuss)

ggplot(data = df_lexdec) +
  aes(x = exp(RT))

Layer 3: adding observations

• we don’t see any observations (i.e., the bars) on the plot, why not?
  • we haven’t told ggplot() how to visualise them
• we have to define a geom: the geometrical object that a plot uses to represent data
  • in ggplot2, geom functions start with geom_
  • we often describe plots in terms of types of geoms they use, e.g., bar charts use bar geoms (geom_bar()), line charts use line geoms (geom_line()), scatterplots use a point geom (geom_point()), etc.

Global and local aesthetics

• in our final plot, the colour of the histograms is slightly transparent
  • we can control this by adding the argument alpha = 0.3 to geom_histogram()
  • alpha takes any other value between 0 and 1

Global and local aesthetics

Customising our plot

• we can improve our axis and legend labels, and also add titles using the labs() function

• let’s also use the function scale_fill_colorblind() from the ggthemes package

  • this creates colourblind-safe colours

• we’ll also use the theme_minimal() function from ggplot2; what does this do?

• try to add the following to your plot

  • change the labels accordingly
  • and add meaningful comments to the code using #

labs(title = "Plot title",
     x = "x-axis label",
     y = "y-axis label") +
  scale_fill_colourblind() +
  theme_minimal()

Commenting

• the code and plot should look something like this:

# histogram of reaction times by native language
ggplot(data = df_lexdec) +
  aes(x = exp(RT), fill = NativeLanguage) + # set aesthetics
  labs(title = "Reaction times by L1",
     x = "Reaction times (ms)") +
  geom_histogram(position = "identity", alpha = 0.3) +
  scale_fill_colorblind() + # make fill colorblind friendly
  theme_minimal() # set plot theme

Saving plots

• we can store plots in our Environment, just like we can store numbers and data as objects
  • you can name objects anything you want
  • but it’s wise to make the name meaningful (e.g., not fig1 or xyz)
• let’s name this plot fig_lexdec_rt, for “figure lexical decision task reaction times”

Barplots

• copy the code for your histogram
• make the following changes to render our barplot
  • remove the name assignment (fig_lexdec_rt)
  • on the x-axis we want NativeLanguage
  • replace geom_histogram() with geom_bar()
    • remove the arguments for the histogram (not position or alpha)
  • change the labels accordingly
  • save the plot as an object with some meaningful name (e.g., fig_lexdec_l1)

• the plot should look something like this:

Combining plots

• one reason to save your plot as an object is so that we can call on it later
  • i.e., you can produce the plot at one point in your document, but decide to only print it in the rendered report lower down
• another reason is so that we can combine multiple plots
  • this can be done with a variety of packages
  • but let’s try it with the patchwork package
  • use + to connect two plots side-by-side
  • or / to present them one on top of the other

Combining plots with +

fig_lexdec_rt + fig_lexdec_l1

Combining plots with /

fig_lexdec_rt / fig_lexdec_l1

6 Exercises

These exercises should be also be included in your script if you upload it to Moodle. Working through the class materials will prepare you for these tasks.

1. Reproduce our histogram as a density plot by replacing geom_histogram() with geom_density()

   • what does this type of plot show?

2. Produce a barplot that shows the number of observations per word class (hint: you’ll need the variable Class from our dataset).

3. Print your density plot and class barplot one on top of the other using the patchwork package

4. Reproduce the following plots as exactly as you can (hint: you will need the position = "dodge" argument):